Productivity measurement and decomposition: An application using panel data

In this paper, an index of technical change is computed for each operating center within a single large firm. The effects of the several determinants of technical change are identified and measured in a decomposition of the technical change index. In addition, the index allows for cross-sectional and inter-temporal comparisons of productivity among centers, and the cross-sectional variation in productivity is found to be large relative to the variation over time. To derive the index of technical change, a panel data set of costs, input prices, and outputs is used to estimate a translog variable cost function. Variable factor productivity and total factor productivity are measured using cost elasticities estimated by the econometric model. The measured productivity indexes are compared to the econometrically estimated indexes to demonstrate the ability of the model specification to accurately represent technical change. The nature of the data set allows comparisons of productivity to be made free from aggregation and non-homogeneity bias. The model is estimated using two levels of aggregation in the outputs and two in the specification of the time dummy variables to investigate how aggregation bias impacts the model results. The work performed in this paper demonstrates the value of productivity analysis to private-work performed in this paper demonstrates the value of productivity analysis to private-sector firms who could apply results such as these in managerial decision making.